Downloaded from SAE International by University of New South Wales, Monday, August 27, 2018

SAE TECHNICAL
PAPER SERIES 960633

Mechanism of Soot and NOx Emission

Reduction Using Multiple-injection
in a Diesel Engine

Zhiyu Han, All Uludogan, Gregory J. Hampson, and Rolf D. Reitz

University of Wisconsin-Madison

Reprinted from: Multidimensional Engine Modeling

(SP-1169)

The Engineering Society

For Advancing Mobility International Congress & Exposition
Land Sea Air and Space® Detroit, Michigan
February 26-29, 1996
INTERNATIONAL
400 Commonwealth Drive, Warrendale, PA 15096-0001 U.S.A. Tel: (412)776-4841 Fax:(412)776-5760
 Downloaded from SAE International by University of New South Wales, Monday, August 27, 2018

The appearance of the ISSN code at the bottom of this page indicates SAE's consent
that copies of the paper may be made for personal or internal use of specific clients.
This consent is given on the condition however, that the copier pay a $7.00 per article
copy fee through the Copyright Clearance Center, Inc. Operations Center, 222
Rosewood Drive., MA 01923 for copying beyond that permitted by Sections 107 or 108
of the U.S. Copyright Law. This consent does not extend to other kinds of copying such
as copying for general distribution, for advertising or promotional purposes, for creating
new collective works, or for resale.

SAE routinely stocks printed papers for a period of three years following date of
publication. Direct your orders to SAE Customer Service Department.

Quantity reprint rates can be obtained from the Customer Sales and Satisfaction
Department.

To request permission to reprint a technical paper or permission to use copyrighted

SAE publications in other works, contact the SAE Publications Group.

GLOBAL MOBILITY DATABASE

All SAE papers, standards, and selected
books are abstracted and indexed in the
Global Mobility Database.

No part of this publication may by reproduced in any form, in an electronic retrieval

system or otherwise, without the prior written permission of the publisher.

ISSN 0148-7191
Copyright 1996 Society of Automotive Engineers, Inc.

Positions and opinions advanced in this paper are those of the author(s) and not
necessarily those of SAE. The author is solely responsible for the content of the paper.
A process is available by which discussions will be printed with the paper if it is
published in SAE transactions. For permission to publish this paper in full or in part,
contact the SAE Publications Group.

Persons wishing to submit papers to be considered for presentation or publication

through SAE should send the manuscript or a 300 word abstract of a proposed
manuscript to: Secretary, Engineering Activity Board, SAE.

Printed in USA
 Downloaded from SAE International by University of New South Wales, Monday, August 27, 2018
960633
Mechanism of Soot and NOx Emission Reduction Using
Multiple-injection in a Diesel Engine
Zhiyu Han, AN Uludogan, Gregory J. Hampson, and Rolf D. Reitz
Copyright 1996 Society of Automotive Engineers, Inc.
ABSTRACT
Engine experiments have shown that with high-
pressure multiple injections (two or more injection pulses
per power cycle), the soot-NOx trade-off curves of a diesel
engine can be shifted closer to the origin than those with
the conventional single-pulse injections, reducing both
soot and NOx emissions significantly. In order to
understand the mechanism of emissions reduction,
multidimensional computations were carried out for a
heavy-duty diesel engine with multiple injections. Different
injection schemes were considered, and the predicted
cylinder pressure, heat release rate and soot and NOx
emissions were compared with measured data. Excellent
agreements between predictions and measurements were
achieved after improvements in the models were made.
The improvements include using a RNG k-ε turbulence
model, adopting a new wall heat transfer model and
introducing the nozzle discharge coefficient to account for
the contraction of fuel jet at the nozzle exit. The present
computations confirm that split injection allows significant
soot reduction with out a NOx penalty. Based on the
computations, it is found that multiple injections have a
similar NOx reduction mechanism as single injections with
retarded injection timings. Regarding soot reduction, it is
shown that reduced soot formation is due to the fact that
the soot producing rich regions at the spray tip are not
replenished when the injection is terminated and then
restarted. With split injections, the subsequently injected
fuel burns rapidly and does not contribute significantly to
soot production. The present work also demonstrates the
usefulness of multidimensional modeling of diesel
combustion to reveal combustion mechanisms and to
provide design insights for low emission engines.
EXTENSIVE RESEARCH is in progress to reduce both
nitrogen oxides (NOx) and particulate (soot) emissions
from diesel engines due to environmental concerns. One
of the emission-control strategies is in-cylinder reduction
of pollutant production. It is well known that it is very
difficult to reduce both NOx and soot production
simultaneously during the combustion process. Many
emission-reduction technologies developed so far tend to
87
University of Wisconsin-Madison
increase soot emission while reducing NOx emission, and
vice versa. For example, retarding fuel injection timing can
be effective to reduce NO formation. However, this usually
results in an increase of soot production. On the other
hand, although increasing fuel injection pressure can
decrease soot emissions, it can also cause higher NOx
emissions at the same time [1]*. Recently, it has been
shown experimentally that with high-pressure multiple
injections, the soot-NOx trade-off curves of a diesel
engine can be shifted closer to the origin than those with
single-pulse injections, reducing both soot and NOx
emissions significantly [2-4].
Nehmer and Reitz experimentally investigated the
effect of double-pulse split injection on soot and NOx
emissions using a single-cylinder Caterpillar heavy-duty
diesel engine [2]. They varied the amount of fuel injected
in the first injection pulse from 10 percent to 75 percent of
the total amount of fuel and found that split injection
affected the soot-NOx trade-off. In general, their split-
injection schemes reduced NOx with only a minimal
increase in soot emissions and did not extend the
combustion duration.
Tow et al. [3] continued the study of Nehmer and Reitz
[2] using the same engine, and included different dwells
between injection pulses and triple injection schemes in
their investigation. They found that at high engine load
(75%), particulate could be reduced by a factor of three
with no increase in NOx and only a 2.5% increase in
BSFC compared to a single injection, using a double
injection with a relatively long dwell between injections.
They also found that triple injection could reduce NOx and
soot emissions at both light and high loads. Another
important conclusion of Tow et al. [3] is that the dwell
between injection pulses is very important to control soot
production and there exits an optimum dwell at a
particular engine operating condition. The optimum dwell
of a double-injection was found to be about 10 degree
crank angles at 75% load and 1600 rev/min for their
engine conditions.
* Numbers in brackets designate References at the end of
the paper.
 Downloaded from SAE International by University of New South Wales, Monday, August 27, 2018

Pierpont et al. [4] confirmed that the amount of fuel Briefly, the spray models include a ’blob’ injection
injected in the first pulse affects the particulate (smoke) atomization model, wall-impingement hydrodynamics
level in experiments where the NOx emission level was (rebounding/sliding) and dynamically varying drop drag
held constant. However, the best double injections were coefficients to account for drop distortions from sphericity.
found to also depend on the spray nozzle included angle. The ignition model is based on the Shell ignition model.
For a production injector with a 125° included angle, The combustion model is a laminar-and-turbulent
which results in significant wall impingement on the piston characteristic-time model. These models were applied to
bowl, the best double injections were found to be those simulate single-injection combustion in several diesel
with 50% to 60% of the fuel injected in the first pulse. They engines under a wide range of operating conditions and
also found that with a combination of EGR and multiple good agreements between predictions and
injections, particulate and NOx were simultaneously measurements were obtained [8].
reduced to as low as 0.07 and 2.2 g/bhp-hr, respectively, The turbulence, gas/wall heat transfer and emissions
at 75% load and 1600 rev/min. models are discussed here. The effects of these models
Other multiple injection studies can be also found in on combustion predictions will be addressed in the next
the open literature [5, 6]. The published experimental section.
works indicate that multiple-injection is an effective mean A modified RNG k-ε model [9] was used in this study. It
to control NO and particulate production during the diesel was shown that this model could predict more realistic
combustion process. In general, multiple injections allow large-scale flame structures compared with the k-ε model
the injection timing to be retarded to reduce NOx used in previous work [7]. These structures influenced
emission, while holding the particulate at low levels. Both in-cylinder temperature predictions and the modified
the amount of fuel injected in the first pulse and the dwell model was able to quantitatively improve NOx emission
between pulses are important for an optimum injection prediction for single-injection combustion cases [8]. The
scheme. RNG k-ε model is formulated as [9]
With the application of multiple injection technology,
the goal of improved injection scheme design and better
control of engine combustion is made difficult by the fact
that design variables are added with flexible injectors. It is
thus helpful to simulate the engine processes with the use
of computational models, which can provide detailed
temporal and spatial information of precisely parameter-
controlled injection and combustion processes.
Patterson et al. [7] performed multidimensional
computations of multiple injections using an improved
KIVA code. They tried to reproduce the experimental
results of Nehmer and Reitz [2] and achieved a fair
success. However, the accuracy of their model prediction
deteriorated for double-pulse injections as the amount of where
fuel injected in the second pulse increased. Kong, Han
and Reitz [8] modified the code by including a modified
RNG k-# turbulence model and turbulence boundary
conditions [9]. Predictions of combustion and emissions of
single-injections were shown to be improved significantly
[8]. These successes motivated the application of the
code to multiple injections in the present study.
It is clear that a good model is necessary in order to and
predict engine combustion and emissions accurately.
Accordingly, the submodels used by Kong, Han and Reitz
[8] were implemented together with improved heat
transfer and injection models. The models were first
applied to the experimental results of the double injections
of Nehmer and Reitz [2]. For better understanding of the
formation of NO and soot during multiple-injection
combustion processes, a set of designed single- and
double-injection schemes were computed. Based on the
computational results, a mechanism of emission reduction
using multiple-injection is suggested.
In Eqs. (1)-(3), k and ε are the turbulent kinetic
MODEL FORMULATION energy and its dissipation rate. ρ, u, τ and µ are density,
velocity, stress tensor and effective viscosity, respectively.
The numerical models used are based on me KIVA-II η is the ratio of the turbulent-to-mean-strain time scale. S
code [10] with improvements in turbulence, gas/wall heat is the magnitude of the mean strain. m=0.5, and n=1.4.
transfer, spray, ignition, high-temperature combustion, The C3 term accounts for the non-zero velocity dilatation
NOx and soot submodels. The submodels for spray
breakup, ignition and combustion used in this study are which is closed
those discussed in the work of Kong, Han and Reitz [8].

88
 Downloaded from SAE International by University of New South Wales, Monday, August 27, 2018

based on a rapid distortion analysis [9]. Other model modeled. Based on the above argument, the discharge
constants are Cs=1.5 [10], Cµ=0.0845, C1=1.42, C2=1.68, coefficient of an injector nozzle, CD, is used in the spray
αk= αε=1.39, η0=4.38, and β=0.012 [11]. atomization model in this study. That is

A recently developed temperature wall function model A e = C DA 0 (7)

[12] was used to predict gas/wall convective heat transfer. or
This model is derived from the one-dimensional energy r = C Dro (8)
conservation equation and accounts for the effect of
thermodynamic variations of gas density and the increase and, to satisfy fuel mass flow conservation
of the turbulent Prandtl number in the boundary layer. Due Ve = V0 / CD (9)
to the density variation of the gas, the dimensionless
temperature is found to be proportional to the logarithmic where Ae and Ve are the effective nozzle cross-section
ratio of the flow temperature to the wall temperature area and fuel jet exiting velocity used in the model,
respectively; A0 and r0 are the geometric section area and
instead of to the arithmetic difference of the two
temperatures, which is the case for incompressible flows radius of the nozzle, respectively; r is the effective nozzle
radius corresponding to Ae, and V0 is the jet exiting
[13]. The advantages of this model include the ability to
use a relative coarse grid size near the wall with accurate velocity corresponding to A0. Again, r and Ve are used as
heat transfer predictions. With the use of this model, Han the initial parameters in the atomization model in the
present study. The effect of CD will be discussed later.
and Reitz reproduced the measured heat fluxes in a
homogeneous charge engine [12]. The heat flux at the The current approach to modeling NO production is
wall is calculated as with the Extended Zel’dovich mechanism [7]. When
comparing with experimental NOx data, a constant factor
*
ρc p T In ( T ⁄ T w ) of 1.533 (the ratio of the molecular masses of NO 2 to NO)
q w = ------------------------------------------- (4)
2.1In ( y ) + 2.513
+ is applied to the NO predictions. The Extended Zel’dovich
Mechanism consists of the following equations as
where described by Bowman [16]
* O + N2 <--> NO + N (10)
+ u y
y = -------- (5)
v
* 1⁄2 1⁄2 N + O2 <--> NO + O (11)
u = (Cµ k) (6)
In Eqs. (4) to (6), qw is the heat flux through the wall; cp is N + OH <--> NO + H (12)
heat capacity; T is the gas temperature; Tw is the wall With the partial equilibrium of Eq. (13) for the hydrogen
temperature; y is the distance from the wall and v is radicals,
molecular viscosity. Noticed that the effects of source O + OH <--> O2 + H (13)
terms (such as chemical heat release and unsteadiness)
are not included in Eq. (4) since they were shown to be and a steady state assumption for N, which results from
insignificant [12]. setting d[N]/dt=0 in the rate equations resulting from Eqs.
The ’blob’ injection model of Reitz [14] is used in the (10), (11), and (12), the Extended Zel’dovich mechanism
present study. It is assumed in this model that the injected can be written as a single rate equation for NO, as
drop parcels have the same sizes as the nozzle exit originally put forth by Heywood [17],
diameter. The advantage of this assumption is that
uncertainties related to the effects of the internal nozzle-
passage flow and nozzle geometry on initial disturbances
and the atomization process can be incorporated into just
one model constant. However, this makes it difficult to where K12=(k11/k1b)(k2f/k2b) and the subscripts 1, 2
model injectors with different nozzle geometry. Recent and 3 refer to Eqs. (10), (11) and (12), respectively. O,
experiments have confirmed that processes such as OH, O2 and N2 are assumed to be in local thermodynamic
supercavitation take place at the nozzle exit under normal equilibrium. The rate constants as recommended by
diesel injection conditions and the measured flow velocity Bowman [16] are:
is close to the velocity calculated from the pressure drop
assuming inviscid incompressible flow [15]. This indicates k1f= 7.6x1013 exp( -38,000/T)
that an effective nozzle section area and a corresponding k1b = 1.6x10 13 (15)
effective flow diameter should be used in the computation
instead of the geometric nozzle exit area and diameter. In
this way, the contraction of fuel jet can be included in the k2f = 6.4x109 T exp( -3150/T)
atomization model. Therefore, spray characteristics
k2b = 1.5x10 9 T exp( -19,500/T) (16)
affected by different nozzle contraction effects (i.e.,
different discharge coefficients), which influence
combustion characteristics and emissions, could be

89
 Downloaded from SAE International by University of New South Wales, Monday, August 27, 2018

k3f = 1.0x1014
k3b = 2.0x1014 exp( -23,650/T)
Since a number of other NOx mechanisms are
believed also to be important in spray and diffusion
flames, work is in progress to explore these other
mechanisms, such as non-equilibrium radials, N2O
intermediate, prompt NO, NO2, and turbulent mixing
effects [18].
Hiroyasu [19] developed a simple soot model which
was later applied to multidimensional diesel combustion
by Belardini et al. [20]. The model predicts the production
of soot mass, Ms, by a single-step competition between
the soot mass formation rate, Msf, and the soot mass
oxidation rate, Mso, according to
d ( Ms)
·
--------------- = M· sf – M so
dt
The Arrhenius formation rate is proportional to the fuel
vapor mass, Mfv, as given by
M· sf = KfMfv
where the formation coefficient is a function of pressure
(bar) and temperature (K) according to
Kf = Asf P0.5 exp(-Esf / RT)
The Arrhenius oxidation rate is proportional to the soot
mass,
6
K o = ------------------- W· nsc
n
(20)
ρ s D nom
(17)
where W· nnsc is the NSC soot mass oxidation rate per unit
surface area (gram-carbon/sec-cm2), Dnom is a nominal
spherical soot particle size, taken to be 25 nm, the soot
density ρs is taken to be 2.5 g/cm3. More details of the
NSC oxidation model are given by Hampson and Reitz
[22].
COMBUSTION MODELING OF MULTIPLE-
INJECTION
Engine and Measurements
The KIVA-II model predictions were compared with
experimental data from a single-cylinder version of the
Caterpillar 3406 heavy-duty truck engine. The emissions
data, given in Table 1, were measured by Nehmer and
(18) Reitz [2]. The engine specifications used in the
experiments and computations are listed in Table 2. The
injector used in the experiments was a common-rail
system with six holes and 0.259 mm hole diameters. The
injection pressure was 90 MPa.
(19) The labeling scheme in Table 1 for the split injection
case gives the percent of the fuel injected in the first and
last pulses, and the dwell between two injections. For
instance, 75- 8-25 represents 75 percent fuel injected in
(20) the first pulse, 25 percent in the second, and an eight
crank angle degree dwell between the two injection
pulses. The value of NOx [2] were adjusted to be based
on mass NO2 as mentioned before.

M· so = KoMs (21) The computations used tetradecane (C14H30) as the

and the oxidation coefficient is a function of pressure,
temperature and the oxygen mole fraction, X o2,
Ko = AsoXO2P1.8 exp(-Eso/RT)
The single-step, two process Hiroyasu soot model is
shown schematically below as fuel mass going to soot
mass and soot mass going to oxidation products (e.g. CO)
Fuel ---- Kf -----> Soot --- Ko ---> CO
fuel due to its similar C/H ratio to the used diesel fuel (the
fuel used in the experiments was Amoco Premier #2). The
computational mesh used represents one-sixth of the
engine combustion chamber (i.e., a 60-degree sector) for
(22) computational efficiency, since the injector has six injector
holes. There were 20 cells in the radial direction, 30 cells
in the azimuthal direction and 18 cells in the axial direction
with 5 cells in the squish region at top dead center. This
mesh resolution, seen in Fig. 1, has been found to give
(23) adequately grid-independent results by Han and Reitz [9].

The Hiroyasu soot model has been modified in the Table 1 Measured Emissions data for single and multiple
present study by replacing the Arrhenius global oxidation injections.
rate equation with the experimentally based oxidation rate Case Eqv. Ratio Inj. Timing Soot NOx
of Nagle and Strickland-Constable (NSC) [21]. Based on No. (BTDC) (g/bhp-hr) (g/bhp-hr)
the oxidation experiments of carbon graphite in an O2 (Nominal)
Baseline 0.461 13 0.083 5.72
environment over a range of partial pressures, the
oxidation rate is modeled by two mechanisms whose 10-3-90 0.501 12 0.163 4.26
rates depend on the surface chemistry involving more 10-8-90 0.494 16 0.184 4.02
reactive "A" sites and less reactive "B" sites and the 25-3-75 0.451 14 0.095 4.66
conversion of "A" sites to "B" sites. With the above 25-8-75 0.453 15 0.084 4.88
formulation of the soot mass oxidation rate, the oxidation 50-3-50 0.455 12 0.100 5.40
rate coefficient of Eq. (22) is replaced by the NSC 50-8-50 0.454 13 0.083 5.15
oxidation rate coefficient 75-3-75 0.473 12 0.076 6.02
75-8-25 0.455 12 0.074 5.81

90
 Downloaded from SAE International by University of New South Wales, Monday, August 27, 2018

Table 2 Engine specifications and conditions.

Cylinder bore x stroke (mm) 137.6 x 165.1

Connecting rod length (mm) 261.62
Displacement volume (L) 2.44
Compression ratio 15.1
Number of nozzle 6 x 0.259
orifice x diameter (mm)
Spray angle (from 27.5
cylinder head) (degrees)
Combustion chamber Quiescent
Piston crown Mexican hat
Inlet air pressure (kPa) 184
Inlet air temperature (K) 310
Intake valve closure -147
(deg. BTDC)
Swirl ratio (nominal) 1.0
Engine speed (rev/min) 1600
Fuel Amoco Premier #2
Injection system Common Rail
Injection pressure (MPa) 90

Figure 2 Comparison showing improvements in

Figure 1 Outline of the computational mesh at TDC.
Spray droplets (circles) are superimposed for
reference.
The Effects of Model Formulation
The effects of the various model formulations have
been studied. As discussed before, the present code uses
the modified RNG k-ε model (Eqs. (1)-(2)), the
compressible-flow heat transfer model (Eq. (4)) and the
discharge coefficient of the nozzle (Eqs. (7)-(9)). Previous
work [7] used the standard k-ε model and the KIVA heat
transfer model [10]. The predicted cylinder pressures and
heat release rates by using the present and the previous
model are shown in Fig. 2 for the 50-8-50 split-injection
case listed in Table 1. The zero-d (also called ’measured’)
heat release rate results are calculated by using an in-
house zero-dimensional code and the measured pressure
data. As can be seen the agreement between prediction
and measurement is improved significantly by using the
present model. Particularly, the previous model [7]
underpredicts the third peak of the heat release rate curve
greatly which indicates that the diffusion combustion that
results from the second-pulse injection is not modeled
accurately.
Computations were carried out to assess the extent to
which the individual models contribute to the differences
seen in Fig. 2. The approach used here was to change
one model at
multiple-injection combustion predictions using the
present model.
a time and to keep the others the same. The effect of the
turbulence model is shown in Fig. 3. When the standard
k-ε model is used, the combustion prediction deteriorates,
which is evidenced by the underprediction of the fuel
burning rate of the second injection pulse that is seen in
the heat release rate curves. Due to the slower
combustion of the second injection pulse predicted by
using the standard k-#, combustion extends later into the
expansion stroke, therefore, the predicted cylinder
pressure is higher than the measured one during the later
expansion stroke.
Two gas/wall heat transfer models, namely, the
present model and the KIVA model [10], were tested.
Figure 4 shows the predicted wall heat fluxes using the
two models. The monitoring location is over the piston
bowl at the edge of the bowl on the cylinder-head surface.
As can be seen, the present model predicts as high as 5.6
MW/m2 peak heat flux during combustion, which is in the
range of previous measured values in heavy-duty diesel
engines [17]. The KIVA heat transfer model is based on
Reynolds analogy [10] and does not account for gas
compressibility. This model was shown to significantly
underpredict heat flux in a gasoline engine by Reitz [23].
In the case discussed in Fig. 4, the heat flux predicted by
using this model is also very low. It is known that in
conventional engines heat is transferred into the cylinder
(heating the gas) also in the earlier portion of the
compression stroke, since the gas temperature is lower
than the wall temperature at this time [17]. This heat
transfer

91
 Downloaded from SAE International by University of New South Wales, Monday, August 27, 2018
Figure 3 Effect of turbulence model on combustion
prediction.
Figure 4 Predicted wall heat fluxes. The monitoring
location is over the piston bowl at the edge of the
bowl at the cylinder-head surface in the plane of the
spray.
characteristic is also predicted by the present model.
As can be discerned in Fig. 4, the present model predicts
negative heat flux (heat is transferred into the cylinder) in
the early compression stroke and positive heat flux (heat
is transferred out of the cylinder) after about 50 degrees
before top dead center, while the standard KIVA model
predicts a positive heat flux throughout the cycle, which is
physically incorrect.
92
Figure 5 Effect of gas/wall heat transfer model on
combustion prediction.
The effect of the heat transfer prediction on
combustion modeling is illustrated in Fig. 5. Since the
standard KIVA heat transfer model underpredicts heat
loss, the cylinder pressure (which indicates the engine
work output) is overpredicted due to energy conservation.
On the other hand, the two heat transfer models do not
result in a significant difference in the heat release rate
except that the premixed-burn peak is somewhat higher
using the standard KIVA model. In the present
multidimensional code, heat release is calculated from the
chemical energy released during combustion, therefore,
heat loss through the wall does not directly affect the
combustion calculation. However, it does affect the
evolution of combustion by affecting the gas thermal field.
As discussed before, the discharge coefficient of
injector nozzle, CD, is introduced in the present
computations. In the engine considered, a common rail
injector is used and the nozzle discharge coefficient is set
to be 0.7 based on the measured data of Tow [24]. Its
effect on the combustion prediction is shown in Fig. 6.
With the use of CD=0.7, the spray velocity increases (see
Eq. (9)) which increases the momentum of the spray. As a
result, drop breakup is increased, fuel-air mixing is
enhanced, and better prediction of combustion is
achieved.
 Downloaded from SAE International by University of New South Wales, Monday, August 27, 2018

Figure 7 Single injection predictions. Spray injection

Figure 6 Effect of nozzle discharge coefficient on
timing is 10.5 degrees BTDC. Solid line-measured,
combustion prediction.
dash line-predicted.

Predictions and Comparison with Measurements Table 3 Comparison the predicted and measured NOx
First, single injection cases were predicted to verify the and soot emissions.
accuracy of the model predictions. The baseline case with Case Measured Calculated Measured Calculated
injection timing of 10.5 degrees BTDC was chosen due to No. particulate Soot NOx NOx
its similar injection timing to that used for the split injection (g/bhp-hr) (g/bhp-hr) (g/bhp-hr) (g/bhp-hr)
cases. The heat release rate and the pressure predictions
Baseline 0.083 0.078 5.72 6.28
am compared in Fig. 7 with measured data. The predicted
pressure and heat release rate shows a reasonably good 10-3-90 0.163 0.200 4.26 3.78
agreement. The over-prediction of the heat release rate 10-8-90 0.184 0.190 4.02 4.11
after the premixed-burn period could indicate that the
25-3-75 0.095 0.08 4.66 4.18
initial flame spread rate is overestimated, as also
discussed by Ricart and Reitz [25]. 25-8-75 0.084 0.067 4.88 4.04
The eight different injection schemes indicated in 50-3-50 0.100 0.16 5.40 5.16
Table 1 were simulated next. Figures 8 and 9 show the
50-8-50 0.083 0.07 5.15 4.45
pressure and heat release rate for all 8 split injection
cases, with 3 and 8 crank degree dwells between 75-3-25 0.076 0.078 6.02 5.61
injections, respectively. The results show a very good 75-8-25 0.074 0.069 5.81 5.10
agreement between the predictions and the
measurements except that the premixed-burn is
underpredicted somewhat in the schemes where 10% fuel
is injected during the first injection pulse. Accurate predictions agree reasonably well with the measured
prediction of the heat release rate and pressure is values. In general, the NOx predictions agree with the
required in order to be able to predict the soot and NOx measured data to within 15%, while the soot is predicted
emissions [7]. to within 20% of the measured particulate in most cases.
Soot and NOx predictions were also performed for the The 10-3-90 and 50-3-50 cases over predict the soot by
baseline single injection and the split injection cases. The larger amounts. However, it is fair to say that generally
results are given in Table 3. No model constants were there is a good agreement with the measured values and
changed during the computations. As can be seen in trends.
Table 3, the

93
 Downloaded from SAE International by University of New South Wales, Monday, August 27, 2018

Figure 8 Pressure and heat release rate predictions for split injections. The dwell between the two injection pulses
is three degrees. Solid line--measured, dash line-predicted. From top: 10-3-90, 25-3-75, 50-3-50 and 75-3-25.

94
 Downloaded from SAE International by University of New South Wales, Monday, August 27, 2018

Figure 9 Pressure and heat release rate predictions for split injections. The dwell between the two injection pulses
is eight degrees. Solid line--measured, dash line-predicted. From top: 10-8-90, 25-8-75, 50-8-50 and 75-8-25.

95
 Downloaded from SAE International by University of New South Wales, Monday, August 27, 2018

MECHANISM OF EMISSIONS REDUCTION

The above comparisons of computation and
measurement give confidence in the combustion and
emission prediction capabilities of the present model. In
order to further study the mechanism of emissions
reduction using multiple injections, a set of injection
schemes was designed and computed which included
three single injections and three double injections. The
injection schemes are shown schematically in Fig. 10. In
the nomenclature 75-8-25 (-10), the number in the
brackets indicates the injection starting angle (ATDC).
The same amount of fuel is injected in all the cases
considered. The needle lift is simply modeled by using
one (or two) step functions with the same step height to
keep the same injection velocity in all cases. In all the
double injections, the total injection duration is 28
degrees, and in the first two single injections, the injection
duration is 20 degrees. In the last case, the injection Figure 10 Schematic showing model injection
duration is 8 degrees longer than the other single injection schemes.
cases, and hence the nozzle radius was reduced by
15.5% to keep the same amount of fuel injected. Figure
11 indicates the injected fuel history for each case. With
the use of the designed schemes, the possible effects of
differences in equivalence ratio and injection timing in the
experimental cases (see Table 1) are eliminated. Besides,
the designed schemes are not far from the experimental
ones, hence the computed results can be confidently
used, based on the good agreement with the experiments
discussed before.
The predicted combustion is illustrated in Fig. 12 which
shows the burnt fuel as a function of crank angle. It is
seen that the fuel-burnt history curves of the double
injections are embraced by those of the two single
injections with the 20-degree injection duration. The
double-injections follow the single (-10) case during the
time corresponding to the first injection-pulse and then
become close to the single (-2) case during the time Figure 11 Injected fuel as function of crank angle. The
corresponding to the second injection-pulse. Between the injected fuel is normalized by the total injected fuel.
injection pulses, a transition occurs due to the dwell period
of the injection. The time at which the transition starts
depends on the amount of fuel injected in the first
injection-pulse. These features indicate that the double-
injection combustion is partly the same as, or close to,
either of the single-injections from a macroscopic
viewpoint. However, the combustion process is
complicated by the second injection-pulse after the dwell
between injections. For example, combustion in the 25-8-
75 (-10) case differs from that in the single (-10) case
soon after ignition, which is due to the non-linear effects of
fuel vaporization and mixing. The fuel-burnt curve of the
single (-10, duration 28) case which has a 28-degree
injection duration is also included within those of the two
single-injections with 20-degree injection duration.
However, as can be seen, it does not follow or become
close to any portion of them after ignition.
The computed soot-NOx trade-off is shown in Fig. 13 Figure 12 Fuel burnt histories showing combustion
for the various cases. It is again confirmed that double evolution. ne burnt fuel is normalized by the total
injections are effective at reducing soot and NOx injected fuel.
emissions. For example, the soot emission of the single (-
10) case is reduced by a factor of 4 using the 75-8-25 (- used, the soot emission is increased significantly at the
10) injection while the NO emission is increased very same time that the NO emission is reduced greatly.
slightly, as indicated in Fig. 13. When the 50-8-50 (-10) It is of interest to examine the reasons for these
double injection is used, both the soot and NO emissions observed emissions reductions. In general, the NO
are reduced below those of the single (-10) case. emissions are reduced by using the double injections
However, when the 25-8-75 (-10) double injection is studied here and the

96
 Downloaded from SAE International by University of New South Wales, Monday, August 27, 2018

soot emissions are reduced in some cases. It is

noticed that soot and NO are also reduced relative to the
single (-10, duration 20) case with the use of the single (-
10, duration 28) scheme. Since combustion of this case is
similar to that of the 50-8-50 (-10) double injection as
indicated in Fig. 12, this case results in about the same
level of soot and NO emissions, as seen in Fig. 13.
However, this scheme would require the use of smaller
nozzle hole sizes in applications.
NO REDUCTION MECHANISM - The predicted
accumulated in-cylinder NO-formation history vs. burnt
fuel mass is shown in Fig. 14. As can be seen, the NO
formation history of the 75-8-25 (-10) case is very similar
to that of the single (-10) case. The injection pause does
not affect NO formation significantly and combustion of
the 25% fuel in the second injection pulse only causes a
small increase in NO production. However, the effect of
the injection pause becomes more significant in the 25-8- Figure 13 Computed soot-NO trade-off of the
75 (-10) case. It delays the major portion of the designed injection schemes.
combustion and reduces the NO formation rate
considerably compared with the single (-10) case.
Comparing the 25-8-75 (-10) case with the single (-2)
case in which the injection timing is retarded 8 degrees,
their NO formation rates become similar during the major
portion of the combustion, as also seen in Fig. 14. These
phenomena are closely related to the combustion
evolution histories in Fig. 15 in which the heat release
rates are given. Although combustion in the 25-8-75 (-10)
and single (-2) cases are quite different before 10 degrees
ATDC due to their different injection timings, combustion
of the fuel in the second pulse is delayed by the injection
pause in the former case. As a result, the major portion of
combustion in the 25-8-75 (-10) case becomes similar to
that of the single (-2) case after 10 degrees ATDC.
It is also noticed in Fig. 15 that combustion in the 75-8-
25 (-10) case becomes very different from that of the
single (-10) case after 5 degrees ATDC since this is the
time that the injection is paused. However, the NO Figure 14 NO formation history as the function of
formation histories of the two cases are not very different burnt fuel mass. The burnt fuel is normalized by the
as discussed before in Fig. 14. It is known that NO total injected fuel. The single injection cases have a
formation is very sensitive to the gas temperature during 20-degree duration.
combustion. As can be seen in Fig. 16, which gives the
mixture mass fraction with temperatures greater than
2200 K vs. fuel mass burnt, the evolution of the
high-temperature gas mixture in the 25-8-75 (-10) case is
similar to that of the single (-2) case after 25% of the fuel
is burnt. The greater amount of high-temperature gas
before 25% fuel is burnt in the former case results in the
slightly higher NO level seen in Fig. 14. It is also seen that
the 75-8-25 (-10) and single (-10) cases have almost the
same high-temperature gas histories before 80% of the
fuel is burnt. Hence, they have very similar NO formations.
By comparing Fig. 14 with Fig. 16, it is of interest to notice
that the large gas temperature difference between the
75-8-25 (-10) and single (-10) cases seen in the later
stage of combustion does not affect the NO formation
significantly, while the relatively smaller temperature
differences between the 25-8-75 (-10) and single (-2)
cases in the earlier stage of combustion result in relatively
larger changes in the NO formations. This can be
explained by the fact that the NO chemistry is sensitive to Figure 15 Computed heat release rates for the single
the early combustion details because these combustion and double injection schemes.
products stay at a high temperature for the longest time,
and the combustion

97
 Downloaded from SAE International by University of New South Wales, Monday, August 27, 2018
Figure 16 In-cylinder high-temperature mixture
fractions. The burnt fuel is normalized by the total
injected fuel.
Figure 17 In-cylinder soot production history for
single and double injection schemes.
region is not cooled by the vaporization of the
continuously injected fuel that occurs in the single
injection case. The NO chemistry is effectively frozen after
about 80% of the fuel is burnt in all the cases considered
(see Fig. 14) because the late-burning mixture has a
shorter residence time at a high temperature due to the
piston expansion.
Based on the above discussion, the NO reduction
mechanism of multiple injections is similar to that of
retarding the injection timing. With the use of multiple
injections, combustion of the second-pulse injected fuel is
delayed by the injection pause. When the percentage of
the first-pulse injected fuel is large (75% in Fig. 14), the
NO formation history of the double injection is like that at a
single injection with the same injection timing. The effect
of combustion of the second-pulse injected fuel does not
influence the NO formation significantly. As the
percentage of the fuel in the first injection pulse becomes
small (e.g., 25%), the NO formation rate of the double
injection becomes similar to that of a single injection
retarded with the dwell-time of the double injection. In this
case, combustion of the first-pulse injected fuel has an
important effect on NO production, and it results in more
NO being formed in the earlier combustion stage, and
hence more total NO production.
98
Figure 18 In-cylinder histories of soot formation,
oxidation and net production.
Figure 19 In-cylinder rich mixture volume normalized
by the instantaneous total volume showing the effects
of the second-pulse injection on fuel/air mixing.
SOOT REDUCTION MECHANISM - As indicated in
Fig. 13, some multiple injection schemes can reduce soot
emissions significantly, while some can increase soot
emissions. Figure 17 gives the soot production history
during the combustion process for the various cases. It is
seen that the in-cylinder soot production histories of the
multiple injections are changed significantly from the
original single injection cases. The peak values of in-
cylinder soot from the multiple injections are largely
reduced due to the injection pause, and the net
productions have different values at the end of
combustion. It is known that the net soot production is the
result of the competition between soot formation and soot
oxidation. This is illustrated in Fig. 18 in which the 75-8-25
(-10) and single (-10) cases are compared. It is seen that
the injection pause affects both the soot formation and
oxidation processes and depresses the soot chemistry.
However, the soot formation is reduced more than the
soot oxidation (as indicated by the numbers and arrows in
the figure). Therefore, the formation-oxidation competition
results in a significant reduction of soot production in the
75-8-25 (-10) case (factor of 4). It is expected that the
second-pulse injected spray also
 Downloaded from SAE International by University of New South Wales, Monday, August 27, 2018

Figure 20 Schematic diagram showing soot-reduction mechanisms of split injections. Left: Single injection. Right:
Split injection.

Figure 21 Computed temperature (K) (top) and soot concentration (g/cm3) (middle and bottom) contours in the
plane of the spray axis. Left: Single injection (-10). Right: Split injection 75-8-25 (-10).

enhances fuel-air mixing. This reasoning is supported by
Fig. 19 which shows the change of the in-cylinder gas
volume containing rich mixtures (equivalence ratio greater
than 2.0) as function of the burnt fuel mass. As can be
seen, the amount of rich mixture is reduced significantly
after the injection pause in the 75-8-25 (-10) double
injection case because of the dispersion of the fuel/air
mixture between the injection pulses which is no longer
maintained by the high momentum fuel jet. This process
tends to lean out the mixture. Soot formation is therefore
reduced.
The mechanisms of for the soot reduction using split
injections are illustrated schematically in Fig. 20. Soot is
formed and accumulates in the tip region of the spray jet.
This soot accumulation at the spray tip has also been
observed experimentally by Dec and Espey in an optically
accessible DI diesel engine [26]. In single injection
combustion, the high momentum injected fuel penetrates
to the fuel-rich, relatively low temperature region at the jet
tip and continuously replenishes the rich region, producing
soot. In a split-injection, however, the second-pulse
injected fuel enters into a

99
 Downloaded from SAE International by University of New South Wales, Monday, August 27, 2018

relatively fuel-lean and high-temperature region which is

left over from the combustion of the first pulse. Soot
formation is therefore significantly reduced because the
injected fuel is rapidly consumed by combustion before a
rich soot-producing region can accumulate. This can also
be seen in Fig. 19, where the last 25% of the injection is
seen not to increase the amount of rich mixture in the
chamber. In addition, the soot cloud of the first spray
plume is not replenished with fresh fuel, instead,
continues to oxidize. As a result, the net production of
soot in split-injection combustion can be reduced
substantially, particularly if the dwell between the two
injections is optimized - long enough that the soot
formation region of the first injection is not replenished
with fresh fuel, but short enough that the in-cylinder gas
temperature environment seen by the second pulse
remains high enough to prompt fast combustion, reducing
soot formation.
This mechanism is demonstrated in the computed
results shown in Fig. 21. Although the computed soot Figure 22 Computed soot-NO trade-off showing
distributions are complicated by the gas flow motion and predicted emission reduction using split injections.
wall/spray interactions, it is seen clearly that the high soot Numbers in the figure are the injection timings
region is located in the leading portion of the spray-
induced flow in both the single and split injection cases (ATDC).
(see the middle row of Fig. 21 at 15 deg. ATDC). By this
time the leading edge of the first formed soot cloud has
moved up the bowl and is located on the cylinder head. In
the single injection case, the larger quantity of soot formed
is being oxidized (a larger quantity because this region
was replenished with fresh fuel until the end of the
injection). As a result, the amount of total in-cylinder soot
is reduced by 25 deg. ATDC (left-bottom plot in Fig. 21,
also see Fig. 18). However, in the split injection case, the
second-pulse injected fuel forms a separate burning
region near the nozzle at 15 deg. ATDC, as seen in the
temperature contours of the right-top plot. The fuel-air
mixture becomes relatively lean compared with the single
injection case as was shown in Fig. 19. Since the soot
formation of the second spray plume is greatly reduced,
no appreciable soot is produced in this region at this time.
By 25 deg. ATDC, however, some amount of soot is seen Figure 23 Measured particulate-NOx trade-off of a
to have been produced from the second spray plume Caterpillar engine at 75% load for single and double
deep in the bowl, but its concentration is very low relative injections [3]. Injection timings were varied from -12
to that from the first spray plume. to +1 ATDC.
The above results explain why multiple injections can
improve the soot-NOx trade-off. On the other hand, soot
increased for the 25-8-75 (-10) case. As discussed before, injection timing is used for the two injection schemes. In
the major portion of the combustion process in the 25-8-
75 (-10) case is delayed by the injection pause, thus it is the single injection cases, NO can be reduced with a
expected that the soot production mechanism in this case corresponding penalty of increased soot with retarding
is similar to that of a single injection case with retarded injection timings. This phenomenon is also seen in the 75-
injection timing in which soot emissions increase due to 8-25 split injection case. However, with the combination of
the deteriorated soot oxidation. This is indeed seen by double injections and retarded injection timings,
comparing the 25-8-75 (-10) case with the single (-2) case
in Fig. 17. significant reductions of NO and soot can be achieved
It is expected that further emissions reductions could simultaneously. These predicted resultsare consistent
be obtained if the injection timing was also varied. For with the experimental observations of Tow et al. [3] shown
example, the predicted soot-NO trade-off trend using in Fig. 23 which demonstrates the dramatic reduction of
different injection schemes is illustrated in Fig. 22. It is particulate that has been observed with a 75-10-25 double
clearly seen that the 75-8-25 double injection shifts the injection. It is worth noting that even larger emission
soot-NO trade-off to a lower level of soot emission relative
to that of the single injection with almost no penalty in the reductions have been observed using triple and quadruple
NO emission in the timing range studied (from -14 to +2 injection in combustion with EGR [6].
ATDC) when the same

100
 Downloaded from SAE International by University of New South Wales, Monday, August 27, 2018
SUMMARY AND CONCLUSIONS
Computations were made of diesel engine combustion
with multiple fuel injections. Improved multiple-injection
combustion predictions were obtained using a RNG k-#
turbulence model, a new wall heat transfer model and the
nozzle discharge coefficient in the injection model to
simulate the contraction of fuel jet at the nozzle exit.
It was found that the turbulence model and the nozzle
discharge coefficient affect combustion modeling
significantly. The original KIVA heat transfer model cannot
predict the heat transfer direction change during the
compression stroke ad greatly underpredicts wall heat flux
during combustion. Although the wall heat transfer
calculation seems not to affect heat release prediction
significantly, it does affect the computed cylinder pressure
due to energy conservation.
The present model was validated by comparisons with
different measured multiple injection schemes. The
computed cylinder pressure, heat release rate and soot
and NOx emissions were compared with measured data.
Good agreement was obtained in most cases. However,
some disagreement was found in the premixed-burn
phase for the cases with 10% fuel injected in the first
injection pulse. This indicates more model improvement
may be necessary.
The computations show that a split injection with a
small percentage (e.g., 25%) of fuel in the second
injection pulse can significantly reduce soot production
while not increasing NO formation levels appreciably. It is
confirmed by the computations that soot emission can be
reduced by using split injections, and, split injection also
allows the injection timing to be retarded to reduce NOx
emission. By using a split injection scheme with retarded
injection timings, both soot and NOx can be reduced
simultaneously. These findings reproduce experimental
results which show the benefit of using multiple injections
[2-6].
The mechanism of soot and NO reduction using
multiple injections was studied computationally using a
designed set of model injection schemes. Based on the
computational results, the NO reduction mechanism is
found to be similar to that of a single injection with
retarded injection timing. Regarding soot reduction, soot
formation is reduced after the injection pause between
injection pulses. The reduced soot formation is shown to
be due to the fact that the soot producing rich regions at
the spray tip are no longer replenished. During the dwell
between injection pulses, the mixture becomes leaner.
With multiple injections, multiple soot formation regions
are formed in the combustion chamber, but, since the
subsequent injections take place into a high temperature
environment left from the combustion products of the first
injection, the injected fuel burns more rapidly, soot
formation rates are decreased, and the net soot
production can be reduced dramatically.
Interestingly, it is also found that a single injection with
a longer injection duration gives similar emission
reduction results to that of a 50-8-50 double injection
which has a similar combustion history. However, the
single injection requires the use of smaller nozzle hole
sizes for the same injected quantity. The use of multiple
injections hence gives greatly increased flexibility in
designing injection schemes for emissions reduction.
101
ACKNOWLEDGMENTS
This work was supported by Ford Motor Co.,
Caterpillar Engine Co., DOE/NASA-Lewis and the Army
Research Office. Support for the computations was
provided by Tacom, San Diego Supercomputer Center
and by Cray Research, Inc. The helpful comments of Prof.
Chris Rutland are appreciated. The authors also thank
Mike Chan for providing additional data in Fig. 22.
REFERENCES
1. Pierpont, D. A. and Reitz, R. D., "Effects of
Injection Pressure and Nozzle Geometry on
Emissions and Performance in a D.I Diesel," SAE
Paper 950604, 1995.
2. Nehmer, D. A. and Reitz, R. D., "Measurement of
the Effect of Injection Rate and Split Injections on
Diesel Engine Soot and NOx Emissions," SAE
Paper 940668, 1994.
3. Tow, T., Pierpont, A. and Reitz, R. D., "Reducing
Particulates and NOx Emissions by Using Multiple
Injections in a Heavy Duty D.I. Diesel Engine,"
SAE Paper 940897, 1994.
4. Pierpont, D. A., Montgomery, D. T. and Reitz, R.
D., "Reducing Particulate and NOx Using Multiple
Injections and EGR in a D. I. Diesel," SAE Paper
950217, 1995.
5. Shundoh, S., Komori, M., Tsujimur, K. and
Kobayashi, S., "NOx Reduction from Diesel
Combustion Using Pilot Injection with High
Pressure Fuel Injection," SAE Paper 920461,
1992.
6. Montgomery, D. T., and Reitz, R. D., "Six-mode
Cycle Evaluation of the Effect of EGR and Multiple
Injections on Particulate and NOx Emissions from
a D. I. Diesel Engine," SAE Paper 960316, 1996.
7. Patterson, M. A., Kong, S. -C., Hampson, G. J. and
Reitz, R. D., "Modeling the Effects of Fuel Injection
Characteristics on Diesel Engine Soot and NOx
Emissions," SAE Paper 940523, 1994.
8. Kong, S. -C., Han, Z., and Reitz, R. D., "The
Development and Application of a Diesel Ignition
and Combustion Model for Multidimensional
Engine Simulation," SAE Paper 950278, 1995.
9. Han, Z., and Reitz, R. D., "Turbulence Modeling of
Internal Combustion Engines Using RNG k-ε
models", Comb. Sci. Tech., 106, 4-6, 267, 1995.
10. Amsden, A. A., O’Rourke, P. J. and Butler, T. D.,
"KIVA-II - A Computer Program for Chemically
Reactive Flows with Sprays," Los Alamos National
Labs., LA- 11560- MS, 1989.
 Downloaded from SAE International by University of New South Wales, Monday, August 27, 2018

11. Yakhot, V., and Orszag, S. A. "Renormalization 23. Reitz, R. D., "Assessment of Wall Heat Transfer
Group Analysis of Turbulence. I. Basic Theory." J. Models for Premixed-Charge Engine Combustion
Sci. Comput., 1, 3, 1986. Computations," SAE Paper 910267, 1991.

12. Han, Z., and Reitz, R. D., "A Temperature Wall 24. Tow, C. T., "The Effect of Multiple Pules Injection,
Function Formulation for Variable-Density Injection Rate and Injection Pressure on Particulate
Turbulence Flows with Application to Engine and NOx Emissions from a D.I. Diesel Engine,"
Convective Heat Transfer Modeling," submitted to Master thesis, University of Wisconsin-Madison,
International Journal of Heat and Mass Transfer, 1993.
1995.
25. Ricart, L. M., and Reitz, R. D., "Visualization and
13. Kays, W. M., and Crawfold, M. E., Convective Heat Modeling of Pilot Injection and Combustion in
and Mass Transfer, McGraw-Hill Company, 1980. Diesel Engines," SAE Paper 960833, 1996.

14. Reitz, R. D., "Modeling Atomization Processes in 26. Dec, J. E., and Espey, C., "Ignition and Early Soot
High-Pressure Vaporizing Sprays. Atomization and Formation in a DI Diesel Engine Using Multiple 2-D
Spray Technology," 3, 309, 1987. Imaging Diagnostics," SAE Paper 950456, 1995.

15. Chaves, H., Knapp, M., Kubitzek, A., Obermeier, F.,

and Schneider, T., "Experimental Study of
Cavitation in the Nozzle Hole of Diesel Injectors
Using Transparent Nozzles," SAE Paper 950290,
1995.

16. Bowman, C. T. "Kinetics of Pollutant Formation and

Destruction in Combustion," Prog. Energy
Combust. Sci., 1, 33, 1975.

17. Heywood, J. B., Internal Combustion Engine

Fundamentals, McGraw-Hill Company, 1988.

18. Hampson, G. J., Lui, Y., Han, Z., and Reitz, R. D.,
"Modeling of NOx Emissions with Comparison to
Exhaust Measurements for a Gas Fuel Converted
Heavy-Duty Diesel Engine," in preparation.

19. Hiroyasu, H. and Kadota, T., "Models foe Combustion

and Formation of Nitric Oxide and Soot in DI Diesel
Engines," SAE Paper 760129, 1976.

20. Belarduni, P., Bertoli, C., Ciajolo, A., D’Anna, A. and

Del Giacomo, N., "Three-Dimensional Calculations
with In Cylinder Sampling Valve Data," SAE Paper
922225, 1992.

21. Nagle, J. Strickland-Constable, R. F., "Oxidation of

Carbon between 1000-2000 C," Proc. of the Fifth
Carbon Conf., Volume 1, Pergammon Press, p.
154, 1962.

22. Hampson, G. J. and Reitz, R. D., "Development of

NOx and Soot Models for Multidimensional Diesel
Combustion," accepted by the ASME International
Joint Power Generation Conference, Minneapolis,
MN, 1995.

102